"It's like having 3 associates and a principal analyst shredding your deal at 3 AM... except they're AI, and they find loopholes humans miss."

DealScout is a multi-agent due diligence system that automates the "Top of Funnel" analysis for Venture Capitalists. Instead of passively summarizing pitch decks, DealScout employs a Combative Multi-Agent Architecture where specialized AI agents actively research, cross-examine, and debate the viability of a startup.

1. Ingest: Accepts detailed Pitch Decks OR just Company Name + URL.
2. Research: If minimal input provided, the Research Agent auto-generates a pitch deck from web sources.
3. Analyze: Market, Product, and Traction analysts verify claims via Google Search & Web Scraping.
4. Debate: The three analysts argue in a simulated IC meeting.
5. Critical Questions: A dedicated agent generates hard-hitting questions to reconsider before investing.
6. Verdict: A GP (General Partner) Agent synthesizes everything into a final "Pass/Invest" memo.

DealScout is powered by 7 specialized agents running on Google Gemini 2.5 Flash via LangGraph:

• 🚀 Quick Mode: Just enter company name + URL. AI auto-generates a pitch deck from web research before analysis.
• Live Agent Debate: Watch the Market, Product, and Traction agents fight over the startup's moat and viability.
• Questions to Reconsider: Get 8-12 hard-hitting questions across Market, Product, Traction, and Team categories before making an investment decision.
• Active Fact-Checking: The system doesn't trust the PDF. It googles the competitors and scrapes the landing page.
• Deep Dive Dashboards: Expandable technical, market, and traction reports with raw data sources.
• Visual "VC Terminal" UI: Dark-mode, financial terminal aesthetic built with Streamlit.
• PDF & URL Support: Drag-and-drop a deck or just paste a URL to start diligence.

• Python 3.10+
• A Google API Key (for Gemini 2.5 Flash)
• Serper API Key (for Google Search capabilities)

1. Clone the Repository

   git clone https://github.com/AKMessi/dealscout.git
   cd dealscout

2. Install Dependencies

   pip install -r requirements.txt

3. Configure Environment Variables

   Create a .env file in the root directory:

   GOOGLE_API_KEY=your_google_api_key_here
   SERPER_API_KEY=your_serper_api_key_here
   

4. Run the Application

   streamlit run ui/app.py

dealscout/
├── src/
│   ├── agent.py           # Main agent implementation with tools
│   └── deal_scout.py      # Simplified agent workflow
├── ui/
│   └── app.py             # Streamlit frontend
├── prompts/
│   ├── market_analyst.md
│   ├── product_analyst.md
│   ├── traction_analyst.md
│   ├── debate_moderator.md
│   ├── questions_to_reconsider.md
│   └── gp_synthesizer.md
├── .env                   # API keys (not in git)
├── requirements.txt       # Python dependencies
└── README.md

DealScout supports two input modes:

Don't have a pitch deck? No problem! Just enter:

• Company Name (e.g., "Anthropic")
• Company Website (e.g., "https://anthropic.com")

Our AI Research Agent will:

1. Scrape the company website
2. Search for recent news, funding, and traction data
3. Generate a comprehensive pitch deck automatically
4. Run the full analysis pipeline

Have a detailed pitch deck? Paste it directly:

• Full Pitch Deck Text with sections like Product, Market, Traction, Team
• Company Website (optional, for additional product analysis)

1. Enter Input: Choose Quick Mode or Full Pitch Deck mode
2. Run Analysis: Click "Run Analysis" to start the 7-agent pipeline
3. Review Results:
   • Check the Dashboard for key metrics
   • Read the Agent Debate for conflicting viewpoints
   • View Auto-Generated Pitch (if using Quick Mode)
   • Review the Deep Dives for detailed analysis
   • Questions to Reconsider - Critical questions before investing
   • Check the Investment Verdict for the final recommendation

1. Go to Google AI Studio
2. Create a new API key
3. Add it to your .env file as GOOGLE_API_KEY

1. Go to Serper.dev
2. Sign up and get an API key
3. Add it to your .env file as SERPER_API_KEY

Contributions are welcome! Please feel free to submit a Pull Request.

This project is licensed under the MIT License - see the LICENSE file for details.

This tool is for educational and research purposes. Investment decisions should not be made solely based on AI analysis. Always conduct thorough due diligence and consult with financial advisors.